Nocturnal agitation is a major management problem in Alzheimer's Disease and other dementias. This descriptive study will use systematic, intensive, 24-hour behavioral observations and measurements of illumination and body temperature to investigate a physiological basis for such sundowning in demented patients in a skilled care nursing facility. Our overall hypothesis is that sundowning is physiologically based in decrements in the circadian timing systems of aged, demented patients. Agitation will be defined through use of a well-developed, highly reliable rating scale used by research assistants 96 times per day to determine both verbal and physical components of agitation. Sundowning will be defined operationally as occurring when over 50% of a given patient's agitated behavior observations occur between sunset and sunrise. Observations will be made continuously 24-hours a day over 3 weeks for a total of 1440 observations per patient. Patients will be studied under conditions of both relatively high illumination (summer) and relatively low illumination (winter) with order counterbalanced to control for terminal decline. On the bases of these observations we will: 1) determine the prevalence of sundowning; 2) fully describe the temporal pattern of agitated behavior over the 24 hour day; 3) study apparent seasonal variation in agitation in dementia (more agitation under lower illumination); 4) determine whether some demented patients may become more agitated by bright light exposure. Pilot data to date suggest high statistical power (>.90) for projected sample size. These results will map directly into non-pharmacological, behaviorally based treatments for sundowning in dementia.